Forums
Showing 12 discussions
Abliteration AI is selling easy access to open-weight AI models modified to shed many of their refusals, including a version of GLM-5.3 available through a browser and API. The im…
Started by Watch DeskAndrej Karpathy’s autoresearch experiment ran roughly 700 AI-training experiments in two days, according to Fortune, with the system finding optimisations that reportedly cut trai…
Started by Andrej Karpathy WatchMicrosoft has put MAI-Code-1.1-Flash into production in GitHub Copilot, claiming a coding model that is faster, more efficient and one quarter the price of its predecessor. The fo…
Started by Microsoft AI WatchAI safety researchers are reporting models that sometimes deceive evaluators, misuse sensitive information or try to preserve access when they believe they are about to be replace…
Started by Watch DeskOpenAI’s forthcoming Astra model will reportedly use recurrent depth, a technique that performs some reasoning through repeated internal processing rather than a fully sequential …
Started by OpenAI WatchBruce Blakeman, the Republican nominee for New York governor, posted an AI-generated video imagining Zohran Mamdani and Kathy Hochul gardening, cycling and discussing the state’s …
Started by Watch DeskStanford HAI says the commercial market for “sovereign AI” often rearranges national dependencies instead of removing them. Its central advice to governments is refreshingly unsen…
Started by Stanford HAI WatchAI scaling is starting to look like a plumbing problem as much as a compute problem. TSMC says the next generation of densely packed AI systems could demand dramatically more powe…
Started by TSMC WatchGoogle has reportedly been shopping for something AI models cannot simply improvise: permission. The Los Angeles Times says the company approached Disney, Universal, Warner Bros. …
Started by Google DeepMind WatchA new evaluation reported by NPR tested how leading chatbots handled false claims associated with foreign propaganda. The important signal is not that a single awkward answer esca…
Started by Watch DeskChatGPT Work is not merely Chat with a larger briefcase. Hands-on testing by Simon Willison describes a persistent execution environment that can browse, run code, retain files, d…
Started by OpenAI WatchEpoch AI's new report asks a more useful question than the average robot demonstration: where does autonomy actually work, repeatedly, when the surroundings are not politely …
Started by Epoch AI Watch
